[ieee 2009 american control conference - st. louis, mo, usa (2009.06.10-2009.06.12)] 2009 american...
TRANSCRIPT
Partial Integrated Guidance and Control of
Surface-to-Air Interceptors for High Speed Targets
Radhakant Padhi, Charu Chawla, Priya G. Das and Abhiram Venkatesh
Abstract— An important limitation of the existing IGC al-gorithms, is that they do not explicitly exploit the inherenttime scale separation that exist in aerospace vehicles betweenrotational and translational motions and hence can be inef-fective. To address this issue, a two-loop partial integratedguidance and control (PIGC) scheme has been proposed in thispaper. In this design, the outer loop uses a recently developed,computationally efficient, optimal control formulation named asmodel predictive static programming. It gives the commandedpitch and yaw rates whereas necessary roll-rate command isgenerated from a roll-stabilization loop. The inner loop tracksthe outer loop commands using the Dynamic inversion philos-ophy. Uncommonly, Six-Degree of freedom (Six-DOF) model isused directly in both the loops. This intelligent manipulationpreserves the inherent time scale separation property betweenthe translational and rotational dynamics, and hence overcomesthe deficiency of current IGC designs, while preserving itsbenefits. Comparative studies of PIGC with one loop IGC andconventional three loop design were carried out for engagingincoming high speed target. Simulation studies demonstrate theusefulness of this method.
Keywords- partial integrated guidance and control, highspeed targets, model predictive static programming, dynamicinversion
I. INTRODUCTION
Stringent performance demands are expected for new-
generation interceptors to engage high speed targets, espe-
cially in defense applications. This is a challenging problem
because of several reasons, which include high relative
velocity (and hence very less time availability), high altitude
engagement and hence limited turning capability (due to
reduced dynamic pressure), hit-to-kill capability (i.e. zero
miss distance) demand etc. The trajectory of an interceptor
is typically divided into three segments consisting of lift-
off, mid-course and terminal phases respectively [1]. Out of
these three phases, the terminal homing is a very crucial
phase. In this phase the interceptor locks on to the target
directly (through a seeker) and the guidance and control
algorithms are designed in such a way that the residual errors
of the prior phases get corrected to achieve the minimum
miss distance. However, conventional separated guidance and
control design may not achieve this objective.
Radhakant Padhi, Asst. Professor (Contact author), is with the Departmentof Aerospace Engineering, Indian Institute of Science, Bangalore, India.Email: [email protected]
Charu Chawla and Priya G. Das, Graduate students, are with the De-partment of Aerospace Engineering, Indian Institute of Science, Bangalore,India.
Abhiram Venkatesh, Formerly Project Asst., was with the Department ofAerospace Engineering, Indian Institute of Science, Bangalore, India.
The conventional guidance and control philosophy works
in three loops. In the outermost loop, a point mass engage-
ment model is considered to generate lateral acceleration
(Latax) commands for achieving minimum miss distance.
These latax commands are then converted into equivalent
body rates in an intermediate loop, which are tracked by
the inner most loop [2]. However, whenever a loop structure
is followed, it introduces time lags between the inner and
outer loops. In case of conventional engagements the target
usually moves slowly and hence the time availability is
relatively large. In such a situation this three-loop structure
usually turns out to be sufficient. However, in case of
incoming high speed targets (like ballistic missiles) the time
availability is quite less (especially in the terminal phase).
Both the guidance and autopilot loops are decoupled and
hence it achieves lethality in a limited sense. Moreover,
three loop design may demand higher maneuverability, which
subsequently may lead to control saturation.
To overcome the deficiencies of the conventional three
loop design, integrated guidance and control design (IGC)
ideas have been proposed in the recent literature. In this
design philosophy, attempt is made to achieve the objectives
using the full nonlinear Six-DOF interceptor dynamics in
a single unified (single loop) framework. Techniques like
feedback linearization [2], State Dependent Riccati Equation
(SDRE) technique [3], etc. have been applied in the IGC
framework.
Existing IGC algorithms do not explicitly exploit the
inherent time scale separation that exist in aerospace vehicles
between rotational and translational motions. Hence, such an
algorithm can become ineffective unless the engagement ge-
ometry is close to the collision triangle. From our numerical
experiments on published ideas of IGC, we observed that
the control surface deflections directly respond to the transla-
tional error correction demands, which leads to instability of
the rotational dynamics. A close examination reveals that the
control surface deflections (which are located at the tail of the
interceptor) can create only minor forces, whereas they create
large moments due to a long moment arms from the center
of gravity. Hence, they are very ineffective in translational
error correction directly, whereas they can be very effective
in turning the vehicle. Hence for any guidance and control
algorithm (including IGC) to be successful, one must exploit
the time scale separation that exist in aerospace vehicles
between faster rotational dynamics and slower translational
dynamics.
To address the deficiency of existing IGC designs, a two-
loop partial integrated guidance and control (PIGC) structure
2009 American Control ConferenceHyatt Regency Riverfront, St. Louis, MO, USAJune 10-12, 2009
FrA07.5
978-1-4244-4524-0/09/$25.00 ©2009 AACC 4184
is proposed in this paper. In this design, the commanded
pitch and yaw rates are directly generated from an outer
loop optimal control formulation, which is solved in a com-
putationally efficient manner using the recently-developed
model predictive static programming (MPSP) technique [4].
The necessary roll-rate command is generated from a roll-
stabilization loop. The inner loop generates the necessary
control using the Dynamic inversion (DI) philosophy. Un-
commonly, in both the loops the Six-DOF interceptor model
is used directly. Six-DOF simulation studies have been
carried out in three dimentional environment to demonstrate
the usefulness of this method.
II. SYSTEM MODEL
The whole system is defined by Six-DOF interceptor
model and 3-DOF target model. The equations are derived
in fin and inertial frame.
A. Interceptor Model
A general nonlinear system of the interceptor can be
defined as
X = f (X ,Uδ ) (1)
where
X , [u v w p q r xm ym zm xm ym zm q1 q2 q3 q4 ζ ]T (2)
Uδ , [δr δp δy]T (3)
The elaborated plant model can be written in the following
form:
u
v
w
=
vr−wq− QSCDm
−Q11g
wp−ur + QSCNBNm
− QSCNδδy
m− (Q21+Q31)g√
2
uq− vp+ QSCNANm
− QSCNδδp
m− (Q31−Q21)g√
2
(4)
p
q
r
=
1IXX
(
−QSdCRM−QSdClξ
δr +QSd2CLP
p
2Vm
)
1IYY
(
QSdCMXCGA −QSCNδ(XCPδ
−XCG)δp
−(IXX − IZZ)pr +QSd2CMQ
q
2Vm
)
1IZZ
(
QSdCMXCGB −QSCNδ(XCPδ
−XCG)δy
−(IYY − IXX )pq+QSd2CNR
r
2Vm
)
(5)
q1
q2
q3
q4
=
12(q4 p−q3q+q2r)
12(q3 p−q4q+q1r)
12(−q2 p+q1q+q4r)
12(−q1 p−q2q−q3r)
(6)
ζ = p (7)
where, u,v,w are the interceptor velocity components and
p,q,r are the actual body rates both defined in fin frame.
ζ is the roll angle. q1,q2,q3,q4 are the quaternions which
defines the attitude of the interceptor [1]. xm,ym,zm defines
the position coordinates of the interceptor in the inertial
frame. The position rates in inertial frame are given by
xm
ym
zm
= T−1IB
ub
vb
wb
(8)
In (8) ub,vb,wb are the interceptor velocity components in
body frame and TIB is the time varying transformation matrix
between inertial and body coordinate system, given by
TIB =
Q11 Q12 Q13
Q21 Q22 Q23
Q31 Q32 Q33
(9)
where the elements of the matrix are obtained through
quaternions (q1,q2,q3,q4) [1]. The inertial acceleration of
the interceptor is given by
xm
ym
zm
= T−1IF
u− (vr−wq)v− (wp−ur)w− (uq− vp)
(10)
where TIF = TBF TIB is the transformation matrix from inertial
to fin frame. In this transformation, TBF is the constant
transformation matrix between body frame to fin frame. Fin
frame is achieved by a π4
rotation of body frame along
positive Xb-axis, pointing in the direction of nose of the
interceptor. Some of the parameters used in the model are
Q, the dynamic pressure, Vm, the total interceptor velocity,
S, the reference area, m, the mass of the interceptor, d, the
diameter, g, the acceleration due to gravity and Ixx, Iyy, Izz
the moment of inertia about body axes respectively. The
aerodynamic force and moment coefficients are evaluated in
fin frame. CD is the drag coefficient, CNAN is the pitch and
CNBN is the yaw force coefficient respectively. CNδ is the tail
normal force coefficient per unit δ per pair, CRMis the rolling
moment coefficient, Clξis the roll moment control coefficient
per unit roll deflection, CLP,CMQ
and CNRare damping
derivatives, CMXCGA and CMXCGB are pitching and yawing
moment coefficients respectively. XCPδis the tail moment
arm (with respect to nose), XCG is the axial position of center
of gravity from nose (in meters). The necessary control Uδ
is generated through fin (control surfaces) deflections.
B. Target Model
The 3-DOF model of target in inertial frame based on flat
earth assumptions is used. Target is not maneuvering and
hence follows a straight line flight path.
Vt
ψt
γt
xIt
yIt
zIt
=
−Dmt
−gsin(γt)
0−gcos(γt )
Vt
Vtsin(γt)Vtsin(ψt)cos(γt)Vtcos(γt)cos(ψt)
(11)
where D is the drag (aerodynamic force), mt is the mass
of the target, Vt is the total velocity of the target, γt is the
flight path angle of the target, ψt is the heading angle of the
target and xIt yI
t zIt are the position coordinates of the target
in inertial frame.
4185
C. Design Philosophy of PIGC
In PIGC, the guidance loop uses the translational dynamics
to generate the commanded pitch and yaw rates which
serves the command for the DI loop. The necessary roll-
rate command is generated from a roll-stabilization loop. In
DI loop, the commanded body rates are tracked to generate
the necessary fin deflections which will drive in meeting
the objective of minimum miss distance as shown in Fig.1.
The objective of almost zero miss distance is achieved by
Body RateCommand Generation(MPSP-DI)
Fin Deflection Generation (DI)
Missile
Computation
Target States
u, v, w, p, q1, q
2, q
3, q
4, xrm, yrm, zrm
p, q, r
p*
q* r*
δ δδr p yp*
ζ
Fig. 1. Block Diagram of PIGC scheme
minimizing the relative error in position and velocity in y
and z direction in fin frame with no control authority in X
direction. To visualize the combined effect of guidance and
I
x
y
Z
Interceptor
I
Target
LOS
o o'
ffI
fI
Inertial Frame
x y zI II
Rotating fin frame coordinates
I zI
f ff
m
m
m
y
xI
zI
x y I
x
yz
Fig. 2. Engagement Geometry in Rotating fin frame
control, transition between three reference frames namely,
body frame, inertial frame and fin frame is considered in
the design. The origin of the body coordinate frame is
assumed to be at the interceptor center of gravity. The inertial
coordinates are defined by earth fixed inertial frame. Since
the interceptor seeker defines the target position relative to
the interceptor body coordinates, it is desirable to define
the interceptor and target position in a coordinate system
parallel to the instantaneous interceptor fin axes with the
origin coinciding with the origin of the inertial frame as
shown in Fig.2. The rotating coordinate system is defined
as [x fI , y
fI , z
fI ] and the inertial frame is defined as [xI , yI , zI ].
The interceptor fin frame fixed on the interceptor body is
defined as [xm, ym, zm]. The relative positions of the target
with respect to interceptor in the rotating frame is given by
Xrel , [xrm yrm zrm]T (12)
The relative velocity between interceptor and target is given
by
xrm
yrm
zrm
= TIF
xIt
yIt
zIt
−
u
v
w
−
qzrm − ryrm
rxrm − pzrm
pyrm −qxrm
(13)
Equation (13) is obtained by differentiating (12) such that
derivative of the transformation matrix TIF is given by [5]:
TIF = −ω TIF , ω =
0 −r q
r 0 −p
−q p 0
(14)
where ω is the skew-symmetric matrix.
III. PIGC: MATHEMATICAL DETAILS
This section describes the implementation scheme of the
PIGC.
A. Guidance (Outer Loop): MPSP
1) Generic Theory: In the MPSP technique, one needs
to start from a “guess history” of the control solution. With
the control guess history, obviously the objective will not
meet, and hence, there is a need to improve the solution. In
this section, we present a way to compute an error history
of the control variable, which needs to be subtracted from
the previous history to get an improved control history. A
general nonlinear system is defined as
X = f (X ,U)
The discretized form of the state and output dynamics are
Xk+1 = Fk(Xk,Uk), Yk = h(Xk) (15)
where X ∈ ℜn, U ∈ ℜm, Y ∈ ℜp and k = 1,2, . . . ,N are the
time steps.
Xk+1 = Xk +∆t( f (X ,U)) (16)
However, from (15), we can write the error in state at time
step (k +1) as
dXk+1 =
[
∂Fk
∂Xk
]
dXk +
[
∂Fk
∂Uk
]
dUk (17)
where dXk and dUk are the error of state and control at time
step k respectively. The respective values for∂Fk
∂Xkand
∂Fk
∂Uk
are obtained from the discretized state model of (16). We
aim for the output vector YN → Y ∗N . Expanding YN about
Y ∗N using Taylor series expansion and then using small error
approximation we can write the error in the output, ∆YN ,
YN −Y ∗N as△YN
∼= dYN =[
∂YN
∂XN
]
dXN . The error in output
obtained as:
dYN =
[
∂YN
∂XN
]([
∂FN−1
∂XN−1
]
dXN−1 +
[
∂FN−1
∂UN−1
]
dUN−1
)
(18)
Similarly the error in state at time step (N −1), dXN−1 can
be expanded in terms of dXN−2 and dUN−2 and so on.
Continuing the process until k = 1, we obtain
dYN = B1dU1 +B2dU2 + . . .+BN−1dUN−1 =N−1
∑k=1
BkdUk
(19)
4186
assuming the error in the initial conditions is zero and it
means dX1 = 0. The key point is that the sensitivity matrices
Bk is computed recursively which will save the computational
time enormously [4]. In (19), we have (N −1)m unknowns
and p equations. Usually p << (N−1)m and hence, it is an
under-constrained system of equations. Hence, an additional
objective is met by minimizing the following objective (cost
function) subjected to the constraint in (19)
J =1
2
N−1
∑k=1
(U0k −dUk)
T Rk(U0k −dUk)
Here, U0k , k = 1, . . . ,(N −1) is the previous control history
solution and dUk is the corresponding error in the control
history. Rk is a positive definite matrices which is chosen
judiciously by the control designer. After carrying out the
necessary algebra [7] we get
dUk = −R−1k BT
k A−1λ (dYN −bλ )+U0
k (20)
Hence, the updated control at time step k = 1,2, . . . ,(N −1)is given by
Uk = U0k −dUk (21)
One can refer to the details of MPSP technique in [4]
2) Mathematical Formulation: MPSP starts with the
initial guess history which is generated from Propor-
tional Navigation (PN) law off line. Curve fitting of the
initial guess history obtained through PN law is done
for the present study. It also provides the value for fi-
nal time/time-to-go (t f ). MPSP algorithm works in two
modes, namely the correction and prediction mode. In
the correction mode, 9 states are considered given as
X ,[
v w q1 q2 q3 q4 xrm yrm zrm
]TEqua-
tions (4), (5), (6) and (13) are discretized using Euler
Integration to get the discretized model [6]. Hence, the size
of the matrix∂Fk
∂ Xkobtained in the present work is 9 by 9.
The control for the outer loop formulation is the commanded
pitch and yaw rate U ,[
q∗ r∗]T
. Hence, the size of the
matrix∂Fk
∂Ukobtained is 9 by 2. The output of the MPSP
is YN ,[
yrm zrm yrm zrm
]T. The desired output is
Y ∗N ,
[
0 0 0 0]T
. In the prediction mode 17 states
are considered as stated in (2) to predict the actual plant
model. The convergence is obtained when the relative error
in position and velocity of target and interceptor in yfI and
zfI direction in fin frame as shown in Fig.2 decreases to a
minimum specified value at t f . The fin deflections are treated
as time varying parameters in the correction mode of the
MPSP formulation.
B. Control(Inner Loop) : Dynamic Inversion
The control loop is based on DI which ensures a good
command tracking [8]. The fin deflection serves as the con-
trol for the inner loop and stabilizes the rotational dynamics.
The error in commanded body rates and the actual body
rates will generate fin deflections by enforcing the error
dynamics. The commanded body rates are obtained from the
guidance (MPSP) loop. The output vector for the inner loop
is defined as Y =[
p q r]T
and the control vector is
Uδ =[
δr δp δy
]T. The governing state model for inner
loop contains only the body rates. The DI formulation results
in a square system (no. of outputs same as number of inputs)
[8]. The control explicitly appears in the first order derivative
of the output variable in the inner loop. Thus,
Y = fy(X)+gy(X)U (22)
The goal is Y → Y ∗, where Y ∗ = [p∗ q∗ r∗]T . The error
of tracking is defined as E , Y (t)−Y ∗(t). Enforcing the
first order error dynamics, such that E → 0 as t → ∞. Here,
ki = diag( 1τ ) and τ = Ts/4 is the time constant of the first
order system. It is selected based on the small settling time
(Ts = 0.4sec) due to the fast nature of the rotational dynamics.
Uδ = [gy(X)]−1[−( fy(X)− Y ∗)−K(Y −Y ∗)] (23)
The updated value of fin deflections Uδ is used for prop-
agation of system dynamics. The inner loop assures the
convergence of the body rates to their desired/commanded
values obtained from the guidance loop such that q →q∗ and r → r∗. In this way, the rotational dynamics will
only be responsible for generating the effective control (fin
deflections) and no over coupling with translational dynamics
is accounted which was happening earlier in IGC design.
IV. SIMULATION RESULTS
For our simulation studies, a surface to air interceptor
is considered which is aiming for a high speed threat in
terminal homing. Both the target and interceptor model
equations are simulated by using Runge-Kutta method to
get the required knowledge of the updated plant dynamics
trajectory [6]. The guidance cycle is 12.5msec and the state
update cycle is 2.5msec. The time of engagement (t f ) is
3.9973sec. Assuming the lethal range of 5m, we got the final
miss distance less than 0.5m. Results appears to be promising
in context with the zero effort miss (ZEM) which is around
500m. It is the distance by which the interceptor will miss
the target if it continues along its present course with no
control applied by the interceptor.
A. Comparison between Three-Loop (Conventional) Design
and PIGC
In conventional systems, guidance and control subsystems
are designed separately and then integrated together into the
missile. The performance of such systems are not optimized
truely. The logic of PIGC is validated by carrying out the
comparative study with the conventional design. We can
observe from the Fig. 3, that the commanded body rates has
a straight line profile compared to that of PN law used in
the conventional design. This comes with the advantage of
natural stabilization of the missile body with faster correction
of miss distance. Figure 4 shows the control fins getting
saturated in case of PN law compared to the case of PIGC. It
shows that the objective of minimum distance with minimum
control effort is met. The control values are much below their
saturation limits. To get the miss distance of the same order,
4187
conventional design uses the maximum fin deflection and
finally goes into saturation as shown in Table.I.
TABLE I
SIMULATION RESULTS FOR DIFFERENT INITIAL CONDITIONS
Zero Effort Miss (m) Miss distance(m) Miss distance(m)Cases (with no control) (Three-Loop Design) (PIGC Design)
1 627.71 2.135 1.0182 558.05 2.655 0.7593 522.80 2.060 0.7484 649.17 0.699 0.5095 692.50 2.341 0.552
0 0.5 1 1.5 2 2.5 3 3.5−5
0
5
Time
roll
ra
te
0 0.5 1 1.5 2 2.5 3 3.5−10
0
10
Time
pit
ch
ra
te
0 0.5 1 1.5 2 2.5 3 3.5−10
0
10
Time
ya
w r
ate
PIGC−MPSP
Conventional (three loop)
Fig. 3. Comparison between conventional three loop design and PIGC
0 0.5 1 1.5 2 2.5 3 3.5−2
−1
0
1
Time
roll
fin
0 0.5 1 1.5 2 2.5 3 3.5−2
0
2
Time
pit
ch
fin
0 0.5 1 1.5 2 2.5 3 3.5−2
0
2
Time
ya
w f
in
Fig. 4. Comparison between conventional three loop design and PIGC
B. Comparison between One-Loop IGC Design and PIGC
In a One-Loop design, the guidance law and autopilot
design are formulated into a single unified state space
framework. A comparative study has also been carried out
between State dependent Ricatti Equation (SDRE) based one
loop IGC design and PIGC approach[3]. Table II shows
the comparative study of one loop IGC and PIGC in terms
of miss distance and zero effort miss for different cases.
It is evident from Table II that the engagement in PIGC
takes place at a miss distance of the order of less than
0.2m. All the results are normalized trajectories. Figure 5
shows the 3D engagement scenario in inertial frame of
PIGC scheme. Figure 6 represents the magnified view of
1
1.5
2
2.5
1
1.5
2
2.51
1.2
1.4
1.6
1.8
DownrangeCrossrange
He
igh
t
M
T
Fig. 5. Engagement in 3D space for PIGC.
0.570.58
0.590.6
0.61
0.195
0.2
0.205
0.21
0.2151.21
1.22
1.23
1.24
1.25
1.26
Down rangeCross range
He
igh
t
Fig. 6. Magnified view in 3D space for One-loop IGC
3D engagement geometry in case of one loop IGC. It is also
seen in Table II that the objective of minimum distance is not
met. Figure 7 shows the comparative plot of body rates of
conventional IGC and PIGC scheme. Figure 8 represents the
comparative plot of fin deflections obtained through SDRE
and DI controller in PIGC scheme.
TABLE II
SIMULATION RESULTS FOR DIFFERENT INITIAL CONDITIONS
Zero Effort Miss(m) Miss Distance(m) Miss Distance(m)Cases (with no control) (One-Loop IGC) (PIGC)
1 55.27 39.767 0.03482 53.17 32.147 0.17143 27.05 11.88 0.24814 50.49 24.85 0.10975 44.82 14.56 0.2167
C. Simulation Study with Different Initial Condition
Different cases were studied with interceptor and target
initial state perturbations to check the effectiveness of the
PIGC scheme. Body rates were perturbed differently, with
roll rate (p) between ±20o, pitch rate (q) and yaw rate (r)
between ±10o. Random deviation of 2% is considered for
both interceptor and target initial conditions together. Figure
9 shows the family of trajectories of 3D engagement. Figures
10 and 11 represents the body rates and fin deflections of the
interceptor.
4188
0 0.5 1 1.5 2 2.5 3 3.5 4−20
0
20
Time (sec)
Roll
rate
0 0.5 1 1.5 2 2.5 3 3.5 4−5
0
5
Time (sec)
Pitc
h r
ate
0 0.5 1 1.5 2 2.5 3 3.5 4−5
0
5
Time (sec)
Yaw
rate
PIGC
One−Loop IGC
Fig. 7. Body rates
0 0.5 1 1.5 2 2.5 3 3.5−2
0
2
Time (sec)
Ro
ll f
in
0 0.5 1 1.5 2 2.5 3 3.5−1
−0.5
0
Time (sec)
Pitch
fin
0 0.5 1 1.5 2 2.5 3 3.5−2
0
2
Time (sec)
Ya
w f
in
PIGC
One−Loop IGC
Fig. 8. Control deflections
0.51
1.52
2.5
1
1.5
2
2.51
1.2
1.4
1.6
1.8
DownrangeCrossrange
He
igh
t
Fig. 9. Engagement Geometry from various initial conditions
0 0.5 1 1.5 2 2.5 3−5
0
5
Time
roll
ra
te
0 0.5 1 1.5 2 2.5 3−5
0
5
Time
pit
ch
ra
te
0 0.5 1 1.5 2 2.5 3−5
0
5
Time
ya
w r
ate
Fig. 10. Body rates of the interceptor
0 0.5 1 1.5 2 2.5 3−2
0
2
Time
roll
fin
0 0.5 1 1.5 2 2.5 3−5
0
5
Time
pit
ch
fin
0 0.5 1 1.5 2 2.5 3−2
0
2
Time
yaw
fin
Fig. 11. Fin Deflection of the interceptor
V. CONCLUSION
Using powerful nonlinear and optimal control design
methods like MPSP and dynamic inversion, this paper
proposes a new philosophy of partial integrated guidance
and control scheme for high speed targets in the terminal
homing loop of interceptors. The newly proposed PIGC
algorithm yields smaller miss distance with minimum control
effort. PIGC scheme uses Six-DOF nonlinear model for
both guidance and control loop. This intelligent manipu-
lation preserves the inherent time scale separation prop-
erty between the translational and rotational dynamics, and
hence overcomes the deficiency of current IGC designs.
However, it preserves the benefits of the IGC philosophy.
Comparative studies of the proposed PIGC scheme with
three loop conventional design and one loop IGC has also
been performed. Simulations with different initial conditions
shows the advantages of the PIGC scheme. The final miss
distance is well within the lethal range.
REFERENCES
[1] G. M. Siouris, “Missile Guidance and Control Systems”, Springer, IncNetLibrary 2004.
[2] P. K. Mennon, E. J. Ohlmeyer, “Nonlinear Integrated Guidance-ControlLaws for Homing missiles”, AAIA Guidance, Navigation and ControlConference and Exhibit 6-9 August 2001, Montreal, Canada.
[3] N. F. Palumbo, T. D. Jackson, “Integrated Guidance and Control: AState Dependent Ricatti Differential Equation Approach”, IEEE Interna-tional Conference on control Applications, Piscataway, NJ, 1999.
[4] R. Padhi and M. Kothari, “Model Predictive Static Programming: AComputaionally Efficient Technique for suboptimal control Design”,International Journal of Innovative Computing,Information and Control,Vol.5, No.2, February 2009.
[5] H. Schaub and J. L. Junkins, “Analytical Mechanics of Space Systems,”AIAA Education Series, 2003.
[6] K. E. Atkinson, “An Introduction to Numerical Analysis,” Jhon Wiley& Sons, 2001.
[7] A. E. Bryson and Y. C. Ho, “Applied Optimal Control”, HemispherePublishing Corporation, 1975.
[8] D. Enns, D. Bugajski, R. Hendrick and G. Stein, “Dynamic Inversion:an evolving methodology for flight control design,” International Journalof Control, Vol.59, No.1, pp. 71-91, 1994.
4189